Design, Implementation, and Characterization of Custom RISC-V Soft-Core Processors for Future Communication Networks

Resumen

In recent decades, new hardware requirements and data processing challenges have arisen with the massive increase of small and portable electronic devices for IoT, Industry 4.0, and nanosatellites. On the one hand, the devices require low-power consumption due to their small form factor; on the other hand, the onboard data processing must be fast enough to attend to the tasks without significant delay in communication. However, this is a difficult challenge today since most portable devices only have a general-purpose processor with a limited and small instruction set. Therefore, full-custom HW architectures are still used in complex and critical applications in these portable devices, but at the expense of more logic utilization and power consumption in our design. Taking advantage of the hardware reconfiguration capacity of FPGAs, it is possible to disable the components we are not using in real-time and, in this way, decrease the power consumption of the whole system. Additionally, operational costs can be significantly reduced, and the flexibility of the nodes to adapt to future heterogeneous standards and communication protocols can be improved. Therefore, applications tend to migrate from ASICs due to all the benefits mentioned above. However, ASICs are still better than FPGAs in some aspects, and engineers must make much effort to reduce the gap between them to make the FPGAs feasible to implement architectures in these devices. Therefore, this thesis explores the possibility of executing specific applications and algorithms on top of customized soft-core processors without needing full-custom HW architectures and, simultaneously, meeting the real-time constraints that traditional general-purpose processors cannot afford. In this way, power consumption can keep as low as possible in these portable devices since we are not introducing additional HW for each specific task. Critical contributions and solutions have been made in this thesis to solve all the challenges in future communication networks, identifying the main bottlenecks in applications such as classical cryptography, PQC, and ECCs, which are essential in every small and portable device today.